Anthropic Launches Claude's Built-in Browser as OpenAI Shuts Down Atlas
新智元 · wechat · 2026-08-27
Anthropic has launched a built-in browser for Claude within the Cowork sidebar, enabling the AI to autonomously browse, click, and fill forms without accessing the user's main browser. The feature supports persistent login states via manual cookie imports and employs a three-layer security system to mitigate prompt injection risks, reducing attack success rates to near zero in tests. Simultaneously, the Claude in Chrome extension exited beta and is now available to all paid users with auto-approval capabilities. Anthropic positions the browser as a fallback tool when connectors fail, avoiding direct competition with Chrome. In contrast, OpenAI's standalone AI browser, Atlas, was shut down after 292 days, highlighting differing strategies for AI-browser integration.
More from Safety
- New Hugging Face Incident Details Reveal OAI's Model Capability Underestimation — RebeccaBellan · 2026-08-27
- LLMs Have Gone Rogue and Hacked Companies 17 Times; Anthropic and OpenAI Lead With 8 Each — RebeccaBellan · 2026-08-27
- METR has more AI eval capacity than US civilian government — connoraxiotes · 2026-08-27
- The Guardian video: everyone hates datacentres — but do we really need them? — nordicinst · 2026-08-27
- TechCrunch Recap: Every Time AI Went Rogue and Hacked Companies — TechCrunch AI · 2026-08-27
- A 'parsimonious' alignment fix: Urbit/Bitcoin-style hierarchical identity and auditable capital flows — curious_vii · 2026-08-27